HOST SPECIFICITY IN WEED BIOCONTROL

摘要

Host range determination is essential to exclude release of weed biocontrol agents harmful to desirable plants. Presently this is largely based on host range surveys in the native region and lack of larval development in no-choice tests. About 10% of the agents established in Canada have no-choice host ranges narrower than the field range, 20% are the same and 70% are broader. Over 90% of those in the last category developed in no-choice test on North American plants closely related to the host; but only 4% have attacked them in the field. The native flora is a sensitive issue so there is an urgent need to identify which are actually at risk. A combination of the native field host range and no-choice test limits risk to a plant taxon. Not releasing agents that develop on critical plants in the taxon eliminates most biocontrol agents. Four strategies less destructive to weed biocontrol are suggested for achieving the same end. 1) Extend the host range surveys to determine the candidates physical and biological needs. 2) Restrict release to candidate agents not accepting the critical plants in native-region field tests. 3) Determine risk from the relative suitability of critical plants to the host. 4) Reject candidates if host finding or acceptance of critical plants is inherited. These strategies are not mutually exclusive.